bootneck, on 14 April 2012 - 10:42 AM, said:
Having read the WFA article mentioned earlier I see that even just looking at divisions the SMHL project appears to have collapsed under the sheer weight of data. Can anybody give any examples of any books or articles published using material gathered by this project? I am assuming that Peter Simkin's article on British Divisions during the 100 days in the Paddy Griffith's book mentioned earlier is an example.
If you were trying to rate the effectiveness of particular regiments and their various battalions, would you not also have to factor in individual unit's morale and disciplinary records?
I was involved with SHLM and collected data for a small number of actions involving the 23rd and 39th Divisions. This was back in the early 1990s before the internet had really taken off and before the amount of data that we have today in digital format was around. It would have been much easier with the OH on CD-Rom.
The problem is that there are a huge number of variables that can influence the outcome of an action, as well as several measures of the outcome itself, so there is a lot of data that needs to be collected. For SHLM we had a two forms - one each for attack and defence. The attack form ran to 7 pages and over 100 variables measuring both inputs and outcomes. The approach was based on the Operational Analysis approach used by Colonel Trevor Dupuy, albeit simplified with fewer variables. See here for more: The Dupuy Institute
. For the analysis I did of 23rd Div at Contalmaison on 10 July 1916 I went through the divisional fireplan and calculated how many 18 pdr and 4.5" howitzer shells would have been fired (as per the plan - though in reality it might have been more or less) in order to assess the weight of fire per yard of trench.
The intention would have have been to collect the data - for every divisional action carried out by the BEF on the Western Front - and run some statistical analysis (i.e. multiple regressions, Structured Equation Modelling) to see which inputs were most important, and the relationships between the inputs and the outcomes. [Sorry... I do a lot of stats in my day job.] Given enough data of good quality, we would have been able to generate some good models. Bryn Hammond is on the GWF so hopefully can chip in.
Gary Sheffield's book on the Somme used some of the findings from SHLM. Peter Simkins has written some papers on the BEF in the 100 days using the data. A chap called George Karger did an MA paper at King's having done the SHLM data collection for the 47th Division, but I don't think it has been published.
Assessing battalion performance is harder, simply because so much of a battalion's performance would be determined by the parent formation. For example, artillery, machine guns, trench mortars were brigade or divisional assets, yet could have a huge impact on how an individual battalion performed in action. Divisional performance itself could vary for a given division during the course of the war.
In my opinion, the only sensible approach to assessing the performance of individual battalions would be first to get an assessment at the divisional level (i.e. as per SHLM). Once one has that, one can then collect battalion level data. One needs to be able to control for formation effectiveness so as to be able to get to a fair assessment of a good battalion in a poor division, or vice versa.
It could be done, but it won't be easy and it won't be quick.